Question about FLAC & lossless

[smackyourfupa]

I understand FLAC is lossless. Say for example I have an audio file with a sample rate of 48000 Hz & I re-encode but change the sample rate to 441000. Wouldn't some quality be lost? The same goes for bit per sample. If I start out with 24 & re-encode to 16 or 8, I presume there would be some quality lost as well.

[KarMa]

It's lossless, as long as you output the same Hz and bit depth as the source. In your scenarios quality will be lost and it won't be a truely lossless transcoding but it will losslessly retain all of the newly downsampled audio it's fed.

[smackyourfupa]

Interesting. Thanks for the info.

[Detmek]

Changing samplerate or bitdepth has nothing to do with FLAC being lossless. Those operations are lossy and preceded FLAC compression. FLAC compression will still be lossless compared to input raw data. You are just changing raw data by resamppling or reducing bitdepth.

[Hoser Rob]

I understand FLAC is lossless. Say for example I have an audio file with a sample rate of 48000 Hz & I re-encode but change the sample rate to 441000. Wouldn't some quality be lost?

Yes, you're going to lose quality but, as mentioned, it has nothing to do with the format. You can resample by multiples of 2 ... eg. 192k TO 48k ... with no problems because it's these things work by multiplication & division, and then everything will divide evenly. But otherwise you'll get quantization errors due to numbers not dividing evenly.

I try to avoid audio resampling such as 48K to 44.1K. In WIndows you'd do this with the ASIO or Wasapi driver but you have to forego software mixing, which isn't a problem for me. I always turn off system sounds anyway. Actually I'm a Linux user and it's a bit easier to do this, and in OS X it's easy too.

The same goes for bit per sample. If I start out with 24 & re-encode to 16 or 8, I presume there would be some quality lost as well.

Well, I'll mention a study or 2 ... one from the AES ... that says otherwise:

https://people.xiph.org/~xiphmont/demo/neil-young.html

http://archimago.blogspot.ca/2014/06/24-bit-vs-16-bit-audio-test-part-ii.html

24 bit audio with high sampe rates is very useful for recording because they often use so many processors and pllugins that the low level quantization distortion becomes a problem. But when they downsample the recording to 16 bit the distortion is gone because it's lower in level than the least significant bit in 16 bit.

This is NOT to say I don't think that 24 bit versions of CDs (SACD, HDTracks etc.) aren't worthwhile. They're frakkiing great.

But that has nothing to do with the bit depth or sample rate. When they do those versions they're quite aware that their audiience, unlike the average user, cares about sound quality and has high quality speakers. So they remaster with that in mind, whereas typical CDs are mastrered to sound good in a car stereo.

So what I do with 24 bit recordings is downsample them to 16/44.1 or 16/48 depending on the input sample rate using Audacity, using dithering. And the result just wipes the floor with the CD version. Not just old CDs that were simply vinyl masters they never bothered to remaster for CD (v.common) but newer ones too. I'm talking about stuff I;ve had for almost 50 years and know like the back of my hand.

And of course you save 2X to 6X the disk space.